Summary of work: The BLSA data set of 39 years of experience with the diagnostic tests for diabetes and with outcome variables known to be related to plasma glucose concentration (development of clinical diabetes, coronary heart disease incidence, microalbuminuria, and mortality) is an extensive experience in men and women across the adult age span. Discrepancies between diagnoses based upon fasting glucose and glucose tolerance are of increasing seriousness with advancing age. A detailed analysis of abnormal glucose tolerance in the Baltimore Longitudinal Study of Aging population of men and women has shown that the earliest detectable abnormality usually occurs much earlier with glucose tolerance testing (GTT) than with measurement solely of the fasting glucose concentration (FPG). In men and women, abnormality on the GTT occurs first 71% in men and 81% of the time in women, while the first detectable abnormality occurs first on the FPG in only 8% and 3% respectively. Kaplan-Meier product limit analyses were performed in order to quantify the time lost in identifying the presence of abnormal test results. In essence, the "residence time" in the Impaired States can be expessed as the time required for 50% of the participants to progress to further abnormality. The median time is 18.5 year to progress from Impaired Glucose Tolerance (IGT) to Impaired Fasting Glucose (IFG) and similar median times are required for IGT to progress to Diabetic Glucose Tolerance and for IFG to progress to Diabetic Fasting Glucose. This loss of warning time is remarkably large. Thus, these data support re-institution of the GTT as a screening test for diabetes. The American Diabetes Association report had recommended in 1997 that the GTT be discontinued , but has recently (2002) reversed this recommendation and now approved both the fasting glucose and the glucose tolerance test as screening procedures for diabetes.